 |
 |
Archives of the TeradataForum
Message Posted: Tue, 03 Jul 2007 @ 22:09:27 GMT
Subj: | | Use of surrogate keys in place of dates |
|
From: | | Steve Hager |
My intuition tells me that if I generate surrogate keys to substitute for actual dates in fact tables where I don't need special null handling,
I will pay a heavy performance price in that my partitioning on the date column will be rendered ineffective. For example, I typically create my
partitioning based on week since most queries involve such a time interval. If the partitioning scheme involved the surrogate key, I would not
get rows for adjacent dates in the same partition because I could not partition by anything other than single SK values. Does the "dynamic
partition elimination" feature eliminate this concern?
| |